Recursive Unveils Flint, an AI Model Built Largely by AI Itself

A new open-weight AI coding model called Flint was released today by Recursive, a startup focused on using artificial intelligence to build and improve artificial intelligence. The company describes the tool as a demonstration that the process of recursive self-improvement, which is designed to let AI systems enhance their own capabilities, is moving from theory to practice. Recursive CEO and cofounder Richard Socher said nearly every aspect of the model's pipeline was done and improved upon by AI.

The company used a system it created called the Eureka Machine to automate the development of the model's components. According to Socher, this approach allowed a team of just a handful of researchers to construct a world-class model in only a few months. The Eureka Machine devised a number of innovations during the process, including a new way of training the base model, new model architectures, and code optimized for specific silicon chips. Recursive shared benchmark scores that suggest Flint outperforms other models of similar size and architecture on coding tasks.

AI is already used to help build, test, and improve AI, a process that can speed up the refinement of a model and testing new ideas with less input from humans. It is also possible to create a loop in which AI decides what improvements to make and then begins another round of improvement. This technique is currently limited, but if AI can contribute to every aspect of model development, the process could theoretically continue indefinitely.

Some experts worry that allowing AI to improve itself in this manner could lead to disaster. These doomsayers claim that AI capable of improving indefinitely could escape human control. These concerns gained wider notice this September when a researcher quit Anthropic and warned that advances in AI, including recursive self-improvement, could lead to human extinction within a few years. The idea of RSI has become controversial, and several bills in Congress propose to ban it.

Recursive sees those fears as more science fiction than reality and believes legislation to prevent RSI is an unnecessary impediment to progress. Socher says Flint was built using a system incapable of escaping human control. He says the goal is for the Eureka Machine itself to achieve full recursive self-improvement, but this would still be overseen by humans who could stop the process if it goes awry. "We hope that this shows people that indeed we have all the steps in place" to build such a system, he says.

Recursive has raised $650 million at a $4.65 billion valuation from investors including Google Ventures, Greycroft, Nvidia, and AMD Ventures. Socher was previously the chief scientist at Salesforce. His cofounders include Jeff Clune, a professor at the University of British Columbia who researches ways of building AI systems capable of learning indefinitely.

Testing showed Flint is both fast and capable of fulfilling coding requests. The model, which has 40 billion parameters, can run on a consumer laptop with 48 GB of RAM, though it is significantly smaller than the largest models available, which can have several trillions of parameters and require racks of industrial servers. In testing, the model was used to build a command line to-do list and a program that scours the web for research related to recursive self-improvement.

It is unclear how much progress other US labs have made in this area, though some published work suggests fully recursive systems are already feasible. A team from Microsoft, Carnegie Mellon University, and the University of Virginia recently published work showing how an AI coding agent can recursively improve itself. Socher says the approach used to build Flint could make cutting-edge AI more accessible. "It's the only way this very small team can produce a model that's actually better than any other model in its class," he told me. "This will hopefully show people that [recursive self-improvement] is actually quite nice and will benefit a lot of people."

Next steps involve the company's stated goal for the Eureka Machine to achieve full recursive self-improvement. Socher says this would still be overseen by humans who could stop the process if it goes awry. The company believes this will show people that the steps are in place to build such a system.